Over the past decade, a number of proposals have been floated to introduce taxes that penalize firms with excessively high CEO pay. For example, recent federal legislative proposals include the Curtailing Executive Overcompensation Act of 2023, or the Tax Excessive CEO Pay Act of 2021. These proposals have not just been made at the federal level—for example, the city of Portland passed a tax penalty on firms deemed to have excessive CEO pay levels.

The drive to rein in CEO pay is extremely well founded. As we (and others) have documented, CEO pay at the largest public companies has seen stratospheric growth in recent decades. Even after a dip in 2022, CEO pay at the 350 largest public firms, for example, has risen by over 1,200% (not a typo) since 1978. CEOs were already among the highest-paid workers in our society in 1978, and growth since then has led to their pay soaring well above even their putative peers in the top 0.1% of earners. The extreme pay of CEOs is not merely a symbolic issue, it has warped the entire upper end of the labor market and contributed to overall inequality. Reining in this pay is absolutely a smart and necessary public policy priority. But how? And, more specifically considering the proliferation of proposals using the lever of tax policy, how to use taxes to restrain CEO pay? We offer some recommendations below.

If one is looking for tax policy solutions to CEO pay:

• The most direct lever, and the one likely to have the most force, is to simply raise top marginal tax rates on ordinary income—adding in numerous higher tax brackets on income over $1 million, $5 million, $10 million, and even $20 million.
  • If one is also hoping to claw back some of the past outsized gains CEOs have made, complementing these tax increases on ordinary income with higher tax rates on capital income would make sense.
• If policymakers instead want to use company-level taxes (as opposed to individual income taxes) to target CEO pay, two key policy decisions must be made: what type of taxes to use, and how to define excessive CEO pay that triggers a tax increase.
  • The best company-level taxes for imposing penalties on firms that set excessive CEO pay are excise taxes—tailored taxes based on specific behavior. One can pick the indicator or threshold that triggers the excise tax and structure the tax to specifically rise as the threshold rises. So, for example, one could penalize firms $1 in excise taxes for every $1 a CEO’s salary exceeds $1 million.
    • What is key is that the base of any penalty tax be tailored in ways to minimize avoidance. Higher corporate income tax rates rightly target shareholders for punishment of excessive CEO pay, but firms are quite good at avoiding corporate income taxes generally. Gross receipts taxes are less well-targeted at punishing the firm’s owners for excessive rates of CEO pay than excise taxes or corporate income taxes.
  • For defining excessive CEO pay, the ability of firms to game some measures for defining these thresholds should be taken seriously. To make this type of gaming harder, two approaches could be taken for defining the threshold for excessive CEO pay:
    • Use the simple level of CEO pay as a trigger point.
    • If it is highly desired to reference a ratio of CEO pay to typical workers’ pay as the trigger, a ratio based on the firm’s CEO pay relative to the typical earnings of workers economywide—or the typical earnings of workers in the industry of the firm—would be best, particularly when compared with ratios based on the CEO’s pay and the pay of the firm’s own employees. Put simply, CEOs have lots of ability to game the reported pay of their own workers in destructive ways to avoid such taxes.

There are levers besides tax policy that should be on the table for those looking to restrain CEO pay and foster more equal division of firms’ income.

• The most significant levers are labor law reforms that boost the ability of workers to bargain collectively. The evidence is strong that unions both boost pay of typical workers and also restrain excess pay of corporate managers and executives.
• Currently nonbinding “say on pay” shareholder votes could be made much more consequential if the boards of directors lost their annual salaries or even their jobs if shareholders voted down a pay package the board had approved for a CEO. These boards of directors are the only institution with the real power to bring down CEO pay, and they should be strongly incentivized to do this.

Below we say more about each of these points.

Raising top marginal rates on ordinary income could restrain CEO pay

Today’s high level of CEO pay is not based on the value they bring to a firm (see here for a discussion of some of this evidence). The clearest summary data indicating this might be the rise of CEO pay even relative to the other highest-paid workers in society, the top 0.1% (see Figure A). It surely is not the case that CEOs have somehow become twice as valuable as even the rest of the top 0.1% in recent decades.

Instead, high levels of CEO pay are the result of the uniquely strong bargaining power they have as the head of public corporations that are in theory “owned” by millions of dispersed shareholders. Essentially, CEOs are bargaining with a firm’s shareholders (along with other possible stakeholders) over what executive pay should be. CEOs will obviously give this negotiation lots of focused attention and will be extremely well motivated to bargain as hard as possible. The millions of diffuse shareholders will be much harder to organize to provide a useful countervailing force to monitor and restrain CEO pay.

But raising top marginal tax rates does not rely on organizing the shareholders to attack the problem of excessive CEO pay, it instead targets the incentive of CEOs themselves. Pushing for higher pay is not costless for CEOs. Among other issues, high pay often generates public outrage. If CEOs were not able to keep as much of each extra dollar they carved out of negotiations over pay because marginal tax rates were increased, the benefit/cost ratio of ignoring the outrage constraint (or other costs) in favor of maximizing pre-tax pay would change. This is not just theoretical. Piketty, Saez, and Stantcheva (2014) showed that falling top marginal tax rates were strongly associated with higher pre-tax inequality and higher shares of total income claimed by the very rich across advanced economies in recent decades. Figure B below, also drawing on their work, shows a clear link more specifically between low top marginal tax rates and high pay of CEOs across countries. In the 13 countries for which both average CEO pay and top marginal rates are available, the negative relationship between top marginal rates and CEO pay is highly significant.1 In short, raising top marginal tax rates to target excessive CEO pay is very likely to work.

Many will likely be unsatisfied by this recommendation to use higher top marginal tax rates to restrain CEO pay because it does not seem targeted specifically enough at CEOs. It is true that higher top marginal tax rates will hit many people who are not the heads of publicly held corporations. But that’s a feature, not a bug. High CEO pay is not a symbolic issue in regard to overall inequality, but it is also not the only issue. The fact that higher top marginal tax rates would hit hedge fund managers, private equity firm owners, and the highest-paid athletes and entertainers in the world is not a downside to the policy. Further, by introducing many new tax brackets that rise even as incomes go from (say) $15 million to $20 million, the tax will disproportionately hit CEOs of the largest public companies even relative to the other members of the richest slice of the U.S. economy.

Raising tax rates on capital income could claw back some of the past outsized pay packages of CEOs

The annual salaries of CEOs of the largest public companies are so large that even one year of this pay instantly makes the recipients among the wealthiest Americans. To get into the top 1% of wealth-holders in the U.S. requires a bit over $13 million in net worth.2 The average pay of the CEOs of the 350 largest public companies in the U.S. in 2021 was more than double this.

Many CEOs earn this level of salary (or greater) for years and years, accumulating huge amounts of wealth. Higher tax rates on ordinary income will not apply to income generated by this accumulated wealth. For that, one also needs to raise tax rates on capital income—dividends, capital gains, and corporate profits. One example might bear this out. In 2022, Mark Zuckerberg, Elon Musk, and Jeff Bezos all reported earning $0 in pay as CEOs of their respective companies.3 Yet anytime the value of the stock of their companies rises, their net worth rises because of the past stock shares they were granted as salary. Currently, this rise in their net worth is taxed far more lightly than ordinary income. This highly preferential rate on capital income is unwarranted and unfair (as the lion’s share of capital income is held by the already-wealthy). Raising capital income rates would be good for the economy generally, but it would also be effectively a clawback of some of the past excesses of CEO pay.

How to maximize the effectiveness of company-level taxes as a tool to restrain CEO pay

Most existing proposals to use tax policy as a lever to restrain CEO pay focus on company-level taxes rather than individual income taxes. In these proposals, some company-level tax—either the corporate income tax or an excise tax or a gross receipts tax—is raised or imposed if CEO pay surpasses some defined threshold. Two issues arise in these types of schemes: which tax should be raised or imposed, and how should the threshold for excessive CEO pay be defined.

What is the best company-level tax to raise if CEO pay is deemed excessive?

Of the company-level taxes included in current proposals, two of them—excise taxes and corporate income taxes—generally fall on the firm’s shareholders. If one is hoping to make the firm’s owners face a greater incentive to rein in executive pay, this is a sensible target. Given that firms are extremely well-practiced in minimizing their corporate income tax obligations—especially through reducing the taxable base for these taxes—it is likely that relying on corporate income taxes to discipline firm behavior could be of limited effectiveness. Further, corporate income taxes are generally only paid by publicly held firms. If firms decide that any corporate income tax-based penalty is too onerous, they could decide to go private and avoid these taxes. This is eminently possible—the share of business income accounted for by non-public companies has risen substantially in recent decades and the number of firms going from publicly to privately held in the past two years has reached near record highs. Excise taxes can be more straightforward and harder to manipulate by firms, and do not rely on firms having corporate income tax liability.

Several subnational proposals include a gross receipts tax. Gross receipts taxes are paid by firms and are a percentage of a firm’s sales. However, just because the legal obligation to pay is on firms, the economic incidence of the tax makes it generally identical to a sales tax, and hence the tax’s overall economic burden generally falls on some combination of both the firm and its customers.4 The precise split of who ends up bearing the burden of gross receipts or sales taxes depends on the elasticity of demand and supply for the good or service that the firm produces. Because a significant portion of their ultimate burden will fall on customers, these taxes seem less well suited for either changing the behavior of firms’ owners or penalizing firms for making bad decisions in regard to CEO pay.

How should the threshold for excessive CEO pay be defined?

The threshold for excessive CEO pay in nearly all existing proposals to use tax policy to rein in CEO pay is the ratio of the salary of a firm’s CEO to the salary of the typical worker in the firm (usually defined as the median worker—that worker who has a salary higher than half of their peers and lower than half of their peers). This target has lots of symbolic appeal but is probably not the optimal metric for a high-stakes policy parameter. Firms have substantial scope to game this ratio without actually reducing CEO pay or raising the pay of typical workers at their firm. The key method for gaming this ratio is known as workplace fissuring. Fissuring involves spinning off some key tasks and functions outside of the firm’s legal boundaries (see here for the cause and consequences of workplace fissuring).

A well-known example of this involves cafeteria services at elite universities. There was once a time when cafeteria workers at these universities were often directly employed by the university. Now, however, nearly all such cafeteria workers are employed by third-party contracting firms that agree to provide services to the university, but these third-party contracting firms, not the university, are the direct employer of the cafeteria workers. One key reason why some large employers choose to contract out services is precisely to lower wages for some workers while claiming to not have been the decisionmaker responsible for this wage suppression. It is not an accident that lower-wage workers are far more likely to perform tasks that are frequently contracted out.

Importantly, contracting out lower-wage workers leads to an increase in the average (and median) wage of the workers who remain directly employed by the “lead” firm (i.e., the firm that contracts out with third-party service providers). This provides a key way that firms could try to reduce their ratio of CEO pay to typical workers’ pay without reducing CEO pay or giving a raise to their own (remaining) employees. Given that workplace fissuring has been widespread to date even without the huge incentive of being able to avoid large penalties based on high ratios of CEO pay to typical workers’ pay, the scope for this type of gaming could be very large.

Additionally, fissuring is socially undesirable for many reasons, so if firms engage in it at scale in response to CEO pay penalties like this, it would be a bad outcome, and not only for the failure to actually rein in CEO pay.

The level of CEO pay is a better trigger for tax penalties than CEO-to-worker pay ratios

If the threshold for excessive CEO pay was defined by its level in dollars rather than its ratio relative to other workers at the firm, it would be far less gameable. After all, if the goal is to influence CEO pay, why not use CEO pay as the metric being targeted?

If one looks at the evolution of the CEO-to-worker pay ratio in recent decades, it is clearly driven far more by the rapid acceleration of CEO pay (the numerator) and not by any particularly dramatic development in typical workers’ pay (the denominator). This is true even as typical workers’ pay growth has been disappointingly slow even relative to more-reasonable benchmarks than CEO pay. Figure C below shows the level of CEO pay as well as the level of typical workers’ pay—both actual and if it had grown with economywide productivity since 1965. Even typical workers’ pay adjusted for productivity growth looks completely flat in its trajectory compared with the path of CEO pay. Because the vertical scale of this graph needs to be very wide to accommodate the growth in CEO pay over this time, any difference between actual typical workers’ pay and the pay they would have received had it grown with economywide productivity is impossible to see—which is essentially the point here! All of the dramatic action in the relative pay of CEOs to typical workers in recent decades has been on the CEO pay side.

Figure D makes this point another way, by displaying actual CEO-to-typical-worker pay ratios and the ratio that would have prevailed had typical worker pay risen with productivity.5 The latter series does show a less-pronounced rise, but the increase is still dramatic, with the CEO pay ratio rising more than ten-fold since the late 1960s. By 2021, if typical workers’ pay had risen with productivity the CEO-to-typical-worker pay ratio would have reached 292-to-1. This is less than the actual 2021 figure of 431-to-1, but this is still a ratio significantly higher than what actually prevailed for several years in the late 1990s and 2010s, years when the excessive level of CEO pay was already a salient political issue. Further, these counterfactual ratios are still several multiples higher than the thresholds for defining excessive pay that are used in most of the tax-based proposals to rein in CEO pay.

Finally, it is far from obvious why a CEO annual salary of $10 million would be considered socially acceptable at a financial firm whose typical employee makes $200,000 annually but socially toxic at a fast-food company where the typical worker makes $20,000 per year. The vast difference in pay between typical workers at fast-food companies and those at financial firms is not driven simply by management decisions. Instead, most of the difference in pay between workers in these industries is driven by a wider market that even CEOs are not free to ignore. To be clear, to say that pay is “driven by the wider market” does not mean that pay is fair or good. It just means different policy tools should be used to address this particular issue. A combination of significantly higher minimum wages and financial transactions taxes (FTTs), for example, would be a much more targeted way to address the pay differences between typical workers at fast-food companies and those at financial firms than would any tax policy based on CEO pay ratios.

If one must use ratios—use typical workers’ pay from the wider industry, not the firm itself

If one was determined to hinge tax penalties on ratios of CEO pay to typical workers’ pay, one could use the pay of typical workers in the wider industry the firm is a part of rather than the firm itself. In this case, fissuring off the lowest-paid workers in their own firm will do much less to change industry-level pay and so the CEO-to-worker pay ratio will be far less affected by the firm’s own decisions. This comparison of CEO pay to the pay of typical workers in the wider industry is the approach EPI uses for constructing CEO pay ratios and it would be far less gameable as a trigger for tax penalties.

The need to reduce CEO pay is urgent enough to look for a broader toolkit of policies

The admirable policy drive to restrain CEO pay often falls into the same trap that other policy efforts do: leaning too hard on tax policy. But like many other policy targets, restraining CEO pay is actually amenable to many other policy tools that are not taxes, and which instead seek to directly restructure the market mechanisms that deliver socially destructive outcomes.

Unions are an effective restraint on excessive CEO pay

It is not a coincidence that the stratospheric rise in CEO pay was accompanied by an acceleration in the downward trend of unionization in the United States. The broader correlation between declining unionization and the rise of inequality is well established, and the highest-quality research documents that this relationship is clearly causal, with the decline of unions leading to higher inequality and a higher share of income claimed by households at the top of the income distribution.

We noted earlier that stratospheric CEO pay levels depend in part on CEOs balancing the personal benefits of higher pay versus the costs associated with public outrage about CEO compensation. Unions are egalitarian institutions with resources who are potentially able to boost these “outrage costs” imposed on CEOs by closely monitoring CEO pay and making other stakeholders and economic observers aware of it. This assumption that unions may be able to boost the outrage costs of high CEO pay and hence partially restrain it is backed by high-quality research documenting the significant role of collective bargaining and unionization in restraining excessive managerial pay.

There are many reasons to support policy efforts to boost workers’ ability to form unions and bargain collectively. But a happy knock-on effect of these efforts would be a significant brake on the rise of CEO pay.

Changing the incentives of the ultimate decisionmakers—boards of directors—should be a key policy goal

At best, the company-level tax changes often used in policy proposals to restrain CEO pay work on the incentives of shareholders. By imposing higher corporate income taxes or excise taxes on firms deemed to have excessive CEO pay, returns to shareholders are reduced. This in theory should provide shareholders greater incentive to rein in CEO pay.

However, there is a long academic literature identifying all of the ways that shareholders lack effective power over the top management of companies regardless of their incentives (see a partial discussion of this literature here). The principal problem is that there are too many shareholders and their investment in any one firm is usually too small a share of their overall wealth to devote lots of time to monitoring every aspect of corporate governance at any given firm. This monitoring problem is a recurring theme in corporate governance, and the main way it is supposed to be addressed is by assembling a board of directors who are meant to represent the interests of shareholders. Yet in reality these boards are generally assembled and picked by the CEO who they are supposed to be managing. This allows CEOs to pick board members friendly to them. Further, being a board member is a desirable job. It often pays highly relative to the time commitment demanded. As the boards are assembled by CEOs, and as board seats are valuable, this makes it quite a risk for board members to get on the wrong side of CEOs by, say, questioning the level of the CEO’s pay.

But it is the firm’s board of directors that has the institutional power to set CEO pay. Given this, any policy proposal that boosts these boards’ incentive to push back on excessive pay is highly promising. Economist Dean Baker has identified one key policy lever—making currently nonbinding “say on pay” shareholder votes much higher stakes for board directors. Currently, the broad group of shareholders of a firm are allowed to weigh in on proposed CEO pay packages through advisory votes (these votes are labelled “say on pay”). But, these votes are nonbinding. In 2022, for example, the shareholders of Live Nation voted to disapprove a pay package for the Live Nation CEO. And yet this shareholder vote was overruled by the group that was supposed to be their representatives (the board of directors) and the disapproved pay package went through (making the Live Nation CEO the second highest-paid CEO of the 350 largest firms in the U.S. for 2022).

These “say on pay” votes were part of the Dodd-Frank legislation in 2010. Further legislation could give them a bit more teeth. For example, if a given firm’s board of directors approved a pay package that was later voted down by the broader shareholders on a “say on pay” vote, the directors could be forced to forfeit their pay for a year—or even be removed entirely. For board members looking to hold onto valuable positions, this would introduce a countervailing consideration besides simply keeping the CEO who picked them happy.

As Baker has noted, a very small share (less than 3% generally) of public firm pay packages are voted down in shareholder “say on pay” votes. Introducing consequences for boards of directors who usher through pay packages deemed to be among the absolute worst in the universe of public companies seems like a sensible step.

Some might argue that simply making “say on pay” votes binding would be a more forceful step. Measures in the United Kingdom and France, for example, would have the CEO continue at their old salary in the event of a negative “say on pay” vote. However, the proposal outlined above would penalize the ultimate decisionmakers offering egregious pay packages—the board of directors. Having a constant countervailing incentive to impose discipline on CEO pay-setting is likely more valuable over the long run than simply making “say on pay” votes binding.

Conclusion

Restraining CEO pay levels is too important a goal to attack with just one single policy lever—and particularly if that policy lever is of limited effectiveness. Current pushes to link enterprise-level tax changes to reduction of CEO pay should be refined to reduce the gameability of these policies. Further, broader non-tax measures to reduce CEO pay should become a key part of these efforts.

Notes

1. The average CEO pay referenced in Figure B is from Fernandes et al. (2012). Their sample of CEOs is from a larger universe than the sample analyzed by Bivens and Kandra (2023), which looks only at the 350 largest firms (ranked by sales) in the United States.

2. Author’s analysis based on data from the Survey of Consumer Finances compiled by the Federal Reserve Board of Governors.

3. Author’s analysis of underlying data from Bivens and Kandra (2023). Jeff Bezos was CEO of Amazon for part of 2022.

4. “Incidence” identifies who faces the ultimate economic burden from a change in taxes. The legal obligation to pay sales taxes often falls on customers of goods. However, because the higher post-tax price caused by a sales tax reduces demand for the goods sold by firms, this transfers some of the incidence of the tax onto these firms.

5. The actual CEO-to-worker pay ratio in this figure is slightly different than what is reported in Bivens and Kandra (2023). Figure D in the current paper uses economywide pay for typical workers and compares with CEO pay, while Bivens and Kandra (2023) compare CEO pay with the typical workers in the industries that the firms are located in. Doing it the Bivens and Kandra (2023) way has many advantages but does require making a choice about how to weight each firm’s CEO-to-worker pay ratio. To avoid this extra confusion of discussing weighting schemes in this paper, we just take economywide averages.